4. Analysis of environmental and biological samples by atomic spectroscopic methods

Quality Assurance for Environmental Analysis Quevauviller/Maier/Griepink (editors) © 1995 Elsevier Science B.V. All rights reserved.

63

4. Analysis of environmental and biological samples by atomic spectroscopic methods M. Hoenig and M.F. Guns Institute for Chemical Research, Ministiy of Agriculture, Leuvensesteenweg 17,3080 Tervuren, Belgium

The rapid and constant development of human activities increases the need for analytical chemistry, which enables important decisions to be made daily in different fields such as industry, environment or health. These developments affect directly the quality of our daily life. A few years ago, the analyst's main concern was to perform determinations under the most appropriate conditions and with control of matrix interferences. However, prior to any measurement, there are two fundamentally important stages: sampling and sample preparation, which are too often overlooked in the quality control of environmental and biological analysis. It would be incorrect to say that these two stages were neglected in the past; most analysts were well aware of their importance. Despite significant progress in instrumentation, the quality of the results did not follow the same trend. It appeared necessary to look beyond the instrumentation and it became increasingly obvious that important errors were mostly associated with sample pretreatment stages. Research trends have focused on these critical steps. As a result, analytical chemists are now much better prepared than previously to develop new methods or to control their validity. Sample preparation and development of methods have now became a growing field along with instrumental improvements. This chapter describes the different sample preparation techniques used prior to atomic spectroscopic analysis and gives an overview of interferences likely to occur in the analysis of environmental and biological samples. The use of the principal atomic spectroscopic techniques and their application to environmental analysis will be discussed, with emphasis on problems associated with preparation of different types of environmental samples.

64

4.1

Anai^is by atomic spectroscopic methods

[Ch.4

Pr^aration of samples

Environmental or biological samples may be divided into those which are already in solution or liquid state (water, blood, serum, urine etc.), and solid samples (soils, sediments, plants, animal tissues etc). Solid samples may contain a high proportion of organic matter (plants, animal tissues) or have a more mineral composition (soils, sediments). For routine analysis by spectrochemical techniques samples are required to be in a liquid form and, hence, solid samples must generally be converted into a solution by an adequate dissolution method. 4.1.1 Liquid samples Solutions can generally be introduced directly for analysis and without any prior treatment. The risk of contamination increases with decreasing analyte concentration; all vessels to be used have thus to be cleaned, rinsed and then soaked overnight in 1-10% (v/v) nitric acid. All vessels should be then rinsed in high purity deionized water. In order to minimize analyte losses by adsorption of metal ions on the vessel or on the suspended particles, the collected samples can be stored for a short time in a refrigerator, and for longer periods in a deep freezer. For the same purpose, aqueous solutions are generally acidified (< pH 1.5, nitric acid). 4.1.2 Solid samples Many types of solid environmental samples are passed into an aqueous solution after a dry ashing or a wet digestion. 4.1.2.1 Dry ashing methods Dry oxidation or ashing may be used to remove organic matter from samples. The sample is weighed in a suitable crucible (generally platinum), heated for several hours at 400-5(X) °C in a muffle furnace, and the residue is dissolved in an appropriate acid. The method is simple and large sample series may be treated at the same time. However, the dry ashing procedure cannot be applied if volatile elements (e.g. Hg, As, Se) are to be determined since they may volatilize during the ashing process. In these cases, oxidants may sometimes be used as ashing aids in order to speed-up the ashing and prevent the volatilization of analytes. Commonly used ashing aids are magnesium oxide and magnesium nitrate. Another likely cause of losses during dry ashing is the retention of the analyte by some of the solid matter present in the system. The solid matter available for this reaction is generally the material of the ashing vessel and the constituents of the ash of the sample itself. The choice of the adequate crucible is therefore of prime importance: the most important retention losses have been reported for silica and porcelain vessels, but they vary with many factors. The most appropriate material for dry ashing methods is platinum.

[Ch.4

Analysis by atomic spectroscopic methods

65

4.1.2.2 Wet digestion methods The majority of wet digestion methods involve different combinations of five acids (nitric, sulphuric, perchloric, hydrochloric, hydrofluoric) and hydrogen peroxide. Nitric acid, boiling at about 120 °C,is the most widely used primary oxidant for the destruction of organic matter. It is commonly used in the presence of sulphuric acid, which partially degrades the more resistant material and also serves to raise the boiling point of the mixture, or with perchloric acid, which continues the oxidation after the nitric acid has been removed. Because occasional explosions with perchloric acid may occur, its use is generally avoided. Oxidations with hydrogen peroxide in acid mixtures containing sulphuric acid are based on in situ involved permono sulphuric acid [1]. Combined with the dehydrating action of sulphuric acid, this reagent rapidly degrades many organic materials. Mixtures with hydrochloric acid are used generally for samples with prevailing inorganic matrices, and combinations with hydrofluoric acid are used to decompose silicates, which are insoluble in the other acids. Acid digestions are usually carried out either in glass or Teflon vessels. Pressure dissolution is essentially a wet digestion procedure which is performed under pressure. With this technique, the loss of volatile elements is avoided and the decomposition of more complex matrices is possible. The limiting factor of pressure digestion in "bombs" is the small amount of organic matter which can be treated. In general, the temperatures involved in wet oxidation methods are very much lower than in dry ashing. The volatilization losses or retentions caused by reaction between the analyte and vessel are then much less frequent. However, possible coprecipitations of the analytes with a precipitate formed in the digestion mixture may sometimes occur. The best known example is the coprecipitation of lead on calcium sulphate precipitates formed when a sample high in calcium is digested with a mixture containing sulphuric acid. The most recent method of wet digestion employs microwaves as the energy source. The main advantages of microwave digestion are speed, efficiency of decomposition for difficult to solubilize samples, and the possibility of automation. This technique has been shown to be suitable for trace element determinations in a wide variety of matrices; in some cases, however, care has to be taken when e.g. organic compounds have to be digested [2]. 4.1.2.3 Direct analysis of solids and solid slurried samples There is a growing interest in the determination of elements in solid samples by atomic spectroscopy without carrying out a dissolution step, in order to avoid contamination and losses during preparation of the sample. This approach cannot directly be used in flame AAS (insufficient dissociation of solid particles in the flame with relatively low temperature) and in ICP-AES (physical and chemical interferences due to differences in transport and dissociation phenomena between the solid slurry samples and solutions used for the calibration), but it is particularly convenient for graphite furnace AAS and when only small amounts of sample are available. However, problems may arise because of unrepresentative sub-sampling and enhanced interferences compared with the analysis of solutions.

56

Anatysis by atomic spectroscopic methods

[Ch.4

These and other problems have been discussed in a comprehensive review by Langmyhr and Wibetoe [3]. Several specially designed devices for direct analysis of solid samples using classical AAS have been recently used: graphite cups [4],probes [5],boats [6], specially designed tubes [7] and tube-platform systems [8]. However, direct analysis of solids by electrothermal atomic absorption spectrometry (ETAAS) is initially handicapped due to some restrictive factors not present in the analysis of dissolved samples: 1. 2. 3. 4. 5. 6.

7. 8.

The greatest problem originates from sample heterogeneity which requires a substantial effort to obtain representative sub-samples. The determination of relatively high analyte concentrations is limited by the minimum representative sample mass that can be introduced into the atomizer. One of the major difficulties associated with solid sampling concerns the availability of appropriate calibrant of similar composition to the samples analyzed. For multi-element analysis this technique is particularly time-consuming compared to actual analysis of solutions. The interference effects observed with solid sampling are greater compared to the dissolved samples whose matrix is simplified as a result of the mineralization. The good contact between the analyte and the graphite surface of the furnace, necessary for reproducible heat transfer from platform to analyte and also for the possible analyte reduction by the graphite prior to the atomization, is not achieved as satisfactorily as in the case of solutions. The use of matrix modifiers, often required to ensure the efficiency of platform techniques, is problematic. Sample introduction into the atomizer is less convenient compared to the dissolved sample.

For these reasons, the precision obtained by solid sampling is generally less than that obtained with solution analysis. Nevertheless, many researchers consider that solid sampling facilitates analysis in some specific cases and may lead to consistent results. In 1974, Brady et al [9,10] proposed an interesting method of solid sampling, the dispensing of water-suspended powdered sample into the atomizer, using a micropipette. At the present time, this alternative of introducing solid material as either a suspension or a slurry appears to be the best approach to overcome some of the difficulties associated with the sampling of solids [11-14]. The recent evolution of ETAAS using platforms, autosamplers and adequate signal processing contributes largely to routine applications of this solid sampling alternative. Studies have also been extended to the use of chemical modifiers to minimize the effects of the matrix components [15,16]. The powdered samples are generally suspended in demineralized water and stirred periodically (magnetic stirring or ultrasonic mixing) just before sampling by the autosampler capillary to avoid sedimentation of the particles, which may result in unrepresentative sampling. To overcome sedimentation in the water suspended samples, Littlejohn et al [17] prepared stable slurries with a thickening agent consisting of acrylic acid polymers (Viscalex, Allied Colloids). For the same purpose, Hoenig et al have employed glycerol [16,18].

[Ch.4

Analysis by atomic spectroscopic methods

67

With regard to the representative sampling of the slurry, it is clear that the presence of large particles in the sample is the most critical factor in the analysis. Dispensing of the solid by the autosampler capillary may be biased by the heterogeneity of the slurry. The number of particles contained in very small samples can be easily evduated from the size distribution. For example, for 5 Mg of a typical silty sediment, the number of particles can vary from several dozen for particle sizes between 16 and 32 /xm to several hundred thousands for particle sizes less than 4 /xm. The importance of an intensive grinding of the sample prior the analysis is thus obvious [18]. At present, slurry sampling-ETAAS is generally used for environmental purposes (soils, sediments, atmospheric particles, ground plant samples, lyophilized animal tissues, suspended solids collected in natural waters); some industrial applications of this technique can be also investigated. 4,1,3 Usual procedures 4.1.3.1 Plants Amongst all the procedures known for the mineralization of plant samples, only a method able to dissolve the silica avoids problems due to retention losses of trace elements by the insoluble silica residues; these losses always occur if a dry ashing procedure followed by a simple acid dissolution of the ashes is applied. Thus, the often ignored undesirable effects are eliminated by a more adequate dissolution of the ashes. For this purpose, the "Methode du C.I.I." (Comit6 Inter-Instituts des Techniques Analytiques) ensures acceptable recoveries. In this method the ashing is carried out at 450 °Cand the silica is removed by a hydrofluoric and nitric acid treatment [19]. This method was tested with environmental certified reference materials (CRMs) and proven to be succesful for the determination of major (Ca, K, Mg, Na, P), minor (Fe, Mn) and trace (Cd, Co, Cr, Cu, Mo, Ni, Pb, Sb, Tl, V, Zn) elements; it results essentially in a total decomposition of the sample. Of course, danger of losses by volatilization of certain elements always exists with such a method. The analysis of mercury, selenium and arsenic, for example, cannot be achieved using a dry oxidation procedure. In this case, a wet digestion circumvents this problem as the metals concerned may be dissolved e.g. a mixture of nitric and sulphuric acids combined with hydrogen peroxide has been shown to be suitable for plant analysis [20]. For the determination of mercury in plant material, the addition of nitric acid can be omitted; concentrated sulphuric acid is then used only. 4.1.3.2 Animal tissues Because of the absence of silica in animal tissues, a simple dry ashing procedure followed by a nitric acid and hydrogen peroxide dissolution of the ashes, is generally sufficient. The method was tested with success for a large variety of samples : muscles, organs and fish meal, with satisfactory recoveries for Cd, Co, Cr, Cu, Mn, Ni, Pb and Zn. Wet digestion procedures may also be used, particularly for the determination of volatile elements. For mercury, the mineralization procedure described for plants can be used. In the case of selenium or arsenic, the most appropriate procedure is the high pressure acid digestion performed in teflon lined bombs which, by maintaining strong acids at temperatures well above their boiling point, accelerates the digestion process in comparison to atmospheric pressure digestion in open vessels; this permits a complete dissolution of samples without risking losses of analytes.

68

Analysis by atomic spectroscopic methods

[Ch.4

Many organic materials can be decomposed satisfactorily in digestion bombs but careful attention must be given to the composition of the sample matrix and to possible explosive reactions with the digestion media. In all cases the size of the sample and the amount of oxidant used must be carefully controlled. For nitric acid bomb digestions of organic samples, the mass of sample intake must generally not exceed 0.1 g whereas the amount of concentrated nitric acid added to this charge has to be in the range of 2.5 to 3.0 ml. 4J.3.3 Soils, sediments and particulate matter Simple mineralization procedures are generally used to determine the mobility of elements e,g, in environmental and agricultural studies. These methods, called "strong attacks", use mixtures of strong mineral acids, except hydrofluoric acid. In comparison to the "total attacks" (using mixtures with hydrofluoric acid which ensures dissolution and elimination of silica), recoveries of the strong attacks maybe slightly poorer, particularly for elements difficult to solubilize, associated for example to silicates. For certain analyses where the total analyte content must be known, the direct analysis of the solid sample by ETAAS may be performed (slurry sampling) instead of using a complex mineralization procedure. Good results are obtained only if the slurry is stirred periodically, just before sampling by the autosampler capillary. This mixing may be done using a mini-stirrer or an ultrasonic probe, both operated manually. However, this alternative is not compatible with the concept of a complete automation of the procedure and robotization of the mixing system may be adapted to the spectrometric system [21]. Electrothermal programs and chemical modifiers used for ETAAS slurry analysis are generally similar to those employed for analysis of solutions. The frequency of analysis is the same as with solutions and in any case higher compared to actual solid sampling, which requires sample weighing for each measurement. The rapidity of the whole procedure thus easily allows repetitive analysis. Consequently, the analysis of slurried samples provides additional advantages compared to direct solid sampling: compatibility of analysis with conventional ETAAS devices with the same speed of analysis as for solutions, the possibility to dilute slurries or to dispense variable sample volumes, straightforward addition of chemical modifiers and the possibility of complete automation and sequential multi-element analysis on the same sample. 4.2

Plasma atomic emission spectrometry

4,2,1 Basic principles and instrumentation In plasma atomic emission spectrometry the sample, generally in liquid form, is introduced through a nebulization device (pneumatic or ultrasonic) into a plasma source, where it is evaporated and dissociated into free atoms and ions, and further additional energy is supplied to enable excitation to higher energy states. A plasma is a highly ionized gas; its high temperature and the dissociation of the analyte compounds to atoms and ions and their excitation are produced by collision with other particles, mainly with free electrons. The excited state is unstable and the atom or atomic ion loses its excess energy either by collision with other particles or by radiative transition to a lower energy level. The resulting radiation is called spontaneous emission. The atomic emission spectroscopic (AES) methods are based on these emission spectra, which are very complex compared to the absorption spectra. Therefore, an atomic emission spectrometer requires an optical bank with good resolution and the possibility for background and inter-element correction.

[Ch.4

Analysis by atomic spectroscopic methods

69

The wavelengths of the emission lines are characteristic of the elements present in the plasma source. The detection of radiation at particular wavelengths allows a qualitative analysis of the sample and the measurement of the intensities at these wavelengths gives rise to the quantitative determination of the analytes. Most analytical plasma sources are electrical gas discharges at atmospheric pressure, usually in argon. The temperatures attained in a plasma source are higher than 5000 °Cin the viewing zone. Each spectrometer for sequential (or multi-element) plasma AES measurements is equipped with a monochromator (or polychromator) for adequate wavelength selection and to collect as much light as possible from a selected spectrum area in the radiation source. A complete plasma atomic emission instrument consists of two main units: the signal generator and the signal processor. The signal generator is represented by the sample introduction system (autosampler, pump, desolvation device, nebulizer etc.) and the plasma source. The signal processor comprises the optics and a unit for data acquisition and processing. Various function and parameters of the instrument are now piloted by a microcomputer and the determinations can be carried out automatically according to a preset analytical programme. 4,2,2 Interferences Interference effects in plasma AES comprise transport interferences (generally nebulization), chemical interferences, ionization interferences and spectral interferences. The degree of interference varies from one instrument to another. However, the most significant impediments to the effective use of any emission spectroscopy equipment are spectral interferences. 4.2.2.1 Transport interferences Transport interferences are observed if the amount of the sample nebulized varies considerably as a function of time. They may be caused by matrix salts or organic compounds and solvents of different viscosity, surface tension or density. This may also occur to some extent for solutions with high mineral acid concentrations, particularly when using ultrasonic nebulizers. Moreover, memory effects may occur if long tubes and large vessel surfaces are used in the nebulization and desolvation systems, but these effects can be easily avoided and controlled. 4.2.2.2 Chemical interferences Due to the high plasma temperature, long residence times and the inert atmosphere in the source, chemical interferences caused by an inefficient sample dissociation or by the formation of thermally stable compounds or radicals, are uncommon if an adequate control of main parameters is ensured. 4.2.2.3 Ionization interferences Easily ionizable elements, such as alkali and alkaline earth elements, may alter the intensities of the emission lines of the analyte. This is a serious problem encountered in direct current and microwave induced plasmas (DCP, MIP), but the effects are minor in the case of inductively coupled plasma (ICP).

70

Ana^is by atomic spectroscopic metliods

[Ch.4

4,2.2.4 Spectral interferences Spectral interferences are observed in every emission method. These interferences are most important in ICP because emission lines that may be expected to be weak or nonexistent in other sources such as flames, are quite intense. All spectral interferences originate from line and continuum spectra of atomic and molecular species present in the source and also from the inherent argon spectrum. They can be classified into four principal groups. 1.

2.

3.

4.

4.3

Spectral overlap occurs when the monochromator of the spectrometer is not capable separating the analyte line from the interfering line. In this case, the automated interelement correction can be used by introducing a previously determined interelement correction factor into the analytical program. With instruments offering unlimited selection of analyte wavelength, these problems may also be avoided by selecting an alternative analyte line that does not exhibit the interference. A broadened line wing of a matrix element in the vicinity of the analyte line may cause spectral interference by partial overlapping the analyte line. This interference may be avoided by moving to another interference-free line or in some cases by using adequate background correction. The selection of a background correction technique depends on the shape of the background emission which can be flat, linearly sloped, curved or structured. Spectral continuum interference maybe caused by one of the matrix components emitting a continuum spectrum at the analyte wavelength which may also be overcome by using an alternative line. Spectrometer stray light intensity depends on the efficiency of the optical system used. The stray light effects (always flat-shaped) may be overcome by using adequate background correction. Atomic absorption spectrometry

Introduced by Walsh in 1955 [22], atomic absorption spectrometry (AAS), has seen a more rapid growth than any other analytical technique since its commercial introduction in the sixties. Over the years, knowledge of the technique, optimization of optics and electronics, automation, and elaborate computer systems for handling large numbers of data have made this technique one of the best and most widely used analytical methods for the determination of major and minor elements in agricultural, environmental, biological, geological, industrial etc, samples. Together with ICP optical emission techniques, AAS seems to be replacing the traditional wet chemical methods and classical arc/spark emission devices intensively used in the past. At the present time the analyst is confronted with an increasing demand for achieving a greater sensitivity, reliability and speed in analyzing complex samples. Modern technology has provided new reagents, procedures and instruments. In this section, an attempt has been made to provide the practising analyst or research scientist with a concise, convenient and critical guide through the vast literature of AAS, in a simplified form. Due to the greater complexity of furnace techniques, these are described in somewhat more detail than the already well-documented flame techniques.

[Ch.4

Analysis by atomic spectroscopic methods

71

4.3.1 Basic principles and instrumentation The basic principles of atomic absorption spectrometry can be expressed by three simple statements: (i) all atoms can absorb light, (it) the wavelength at which light is adsorbed is specific for a particular chemical element and (Hi) the amount of light absorbed is proportional to the concentration of absorbing atoms. An atomic absorption spectrometer is composed of: 1. A primary source to generate light at the wavelength which is characteristic of the analyte; 2. An atomization device to create a population of free analyte atoms; 3. An optical system to direct light from the primary source through the atom cloud and into the monochromator; 4. A monochromator to separate light at the analyte wavelength from all other light; 5. A light-sensitive detector and suitable electronics to measure and translate detector response into a useful analytical signal. The principle of operating an atomic absorption spectrometer is that the primary source is used as the resonance line source. This primary source is usually a hollow cathode lamp where the cathode contains the element to be determined. The light beam consists of resonance radiation which is electronically or mechanically pulsed. When a sufficient voltage is applied across the electrodes, the filler gas inside the lamp is ionized and the ions are accelerated towards the cathode. As these ions hit the cathode, they cause the cathode material to sputter and form an atomic vapor in which atoms are present in an excited electronic state. In returning to the ground state, the lines characteristic of the analyte are emitted and passed through the atomization device. Other types of primary sources (EDL lamps or Superlamps with an additional boost-discharge electrode) can be used in some cases, principally for elements determined in the far UV wavelengths (As, Se etc.). The ground state atoms which are produced in the atomizer (usually flame or furnace), and which predominate under the experimental conditions, absorb resonance radiation from the primary source, reducing the intensity of the incident beam. A monochromator isolates the resonance line and allows this radiation to fall on a detector. An electrical signal, whose magnitude depends on the light intensity, is produced. The electronic device is designed to respond selectively to the modulated radiation from the primary source, measure the light attenuation by the sample and convert these readings to the actual sample concentration. Atomic absorption spectrometry is a comparative method. In practice, quantitative analysis is a matter of converting samples and calibrants into solutions, comparing the instrumental responses of calibrants and samples, and using these comparative responses to establish accurate concentration values for the element of interest. Basically, this can be carried out using simple equipment and simple procedures. Inevitably, however, there are aspects of the technique which are not quite as simple and straightforward as this brief introduction suggests. These aspects will be discussed more extensively in the following sections. The use of automated sampling devices and automated recording of results is an effective means of increasing analytical productivity. Automatic sample exchangers usually take the form of a rotating or rectilinear table from which samples are successively presented to a capillary attached to the nebulizer. Similar sample exchangers are also used to present the samples to a graphite furnace. Similarly to spectrometers.

72

Analysis by atomic spectroscopic methods

[Ch.4

sample exchangers have benefited the development of microprocessor technology. Microprocessor- or computer-controlled spectrometers and sample exchangers/dispensers can be set up to interact in a very sophisticated way. The software of modem instrumentation provides extensive possibilities for routine analyses. Generally, a sequence of computer programs are stored on disk. These allow prerecording, storage and recall of analytical methods. The programs include a control of the possible baseline drift, specific algorithms for data manipulation and processing, such as calibration calculations and curve fitting, standard addition regression, averaging, standard deviation, quality control criteria etc. Details of the electrothermal program and both specific and background absorbance time profiles can also be displayed. This represents a very important improvement for method development and research of optimal parameters in furnace analyses. 4,3,2 Flame atomization A successful atomic absorption determination depends on the generation of a supply of uncombined analyte atoms in the ground state and their exposure to light at their specific absorption wavelength. This process consists of taking a solution of the analyte and heating it to a temperature that is sufficient to dissociate the occurring compounds. Usually, the thermal energy required is supplied by a flame. The more sensitive but more problematic electrothermal (furnace) atomic absorption spectrometry is, however, now widely used for trace and ultra-trace analysis. For the atomization, the sample is usually sprayed into the flame in the form of a solution by means of a pneumatic nebulizer. A flame is simple, inexpensive, easy to use, and provides a stable environment for atomic absorption. The complete process includes nebulization of the sample, the selection of mist droplets of the correct size and distribution, the mixing of the selected mist with the flame gases and its introduction into the burner which ensures the atomization. Due to the continuous and stable signal supplied by the flame AAS, the sensitivity of the method is given by the so-called "characteristic concentration", defined as the concentration of the analyte (in iLig.mr^)that produces a 1 % absorption signal (0.0044 absorbance) in the flame. 4,3,2,1 Nebulization In a typical burner-nebulizer system, the sample is transferred into the nebulizer by the low pressure created around the end of the aspiration capillary by the flow of the carrier gas, and passed through the section. The liquid stream is transformed into a droplet spray by a carefully positioned obstacle (usually a spherical glass bead) and is ejected with the carrier gas into the spray chamber. During nebulization some liquids form a finer mist than others. Only about 10 % of the solution is converted into sufficiently fine droplets to be carried into the burner. The droplets with a diameter greater than about 5 jLtm fall out onto the sides of the chamber and flow to waste. The fuel gas is introduced into the chamber along with the carrier gas, and a mixture of fine sample mist, fuel, and carrier gas is transferred from the spray chamber to the burner. Differences in the uptake rates between samples and calibrants will clearly affect analytical accuracy, and nebulization must be identical for all samples and calibrants in a particular analysis. Since the amount of fine mist per unit volume reaching the flame affects the magnitude of the absorption signal, it is important to nebulize samples and calibrants of similar physical characteristics (solvent composition, surface tension, dissolved salts etc.).

[Ch.4

Analysis by atomic spectroscopic methods

73

4.3.2.2 Flames The main function of the flame is to convert the analyte to the atomic state. Considering that the gases in the flame, and thus also the sample constituents, flow with a relatively high velocity, atomization processes should occur as quickly as possible. Measurement of the absorbance should be performed at a position in the flame at which either atomization is complete or equilibrium has been reached. In modem flame atomic absorption spectrometry (FAAS), air-acetylene and nitrous oxide acetylene mixtures are almost universally used for routine analyses. The best known and most often used flame in atomic absorption analysis is the airacetylene flame, generally used with a long path burner. The air-acetylene flame is convenient and safe for working at wavelengths above 200 nm, with minimum chemical interference. Its temperature is about 2300 °C. This flame can be used successfully for the elements that do not form refractory oxides and thus are easily atomized. Temperature is not always the main decomposition factor. Within each flame type the chemical nature of the flame (oxidizing, stoichiometric or reducing) will also have a profound effect on the decomposition behaviour of many elements. Only in a few cases do noticeable ionization interferences occur (alkali metals), with a consequent change in the spectral response. The degree of ionization is different for each element, depending on the energy required to remove electrons. This energy can be supplied in various ways, but for atomic absorption the major source is the heat of the flame. The degree of ionization increases as the analyte concentration decreases; consequently, instead of the more normal calibration graph (nominally linear but in practice curving towards the concentration axis at high absorbance), the curvature tends towards the absorbance axis. This concerns only the ionization in pure solution of the particular element. The presence of any other element with an ionization potential close to or lower than the analyte will modify the extent of ionization significantly. In practice, the effective means to avoid interference due to ionization is to "buffer" the calibrants and samples with a high concentration of an easily ionized element. Provided that the concentration of this element is much greater than the analyte element, and its ionization potential lower, essentially complete suppression of the analyte ionization may be achieved. Due to their low ionization potential, sodium, potassium and cesium are the most commonly used ionization buffers. Nevertheless, the temperature of the air-acetylene flame is insufficient to dissociate a substantial number of principally oxidic bonds or to prevent their formation in the flame. Certainly the most important development in the flames was the introduction of the nitrous oxide-acetylene by Willis in 1965 [23]. As a result of its low burning velocity, this hot flame (2900-3000 °C) offers a favourable chemical, thermal and optical environment for virtually all metals that give difficulties in the air-acetylene flame. From the previous discussions it is evident that the analyst can maximize the population of analyte atoms in the flame by considering the adequate flame type, suitable flame stoichiometry and appropriate solution chemistry. However, before effective analytical measurements can be made it is necessary to ensure that light at the characteristic wavelength passes directly through the analyte atom population. Free analyte atoms are not distributed evenly within the flame envelope, and under given flame conditions there will be a particular zone which is more densely populated by the analyte atoms than other parts of the flame. The location of this zone within the flame is not identical for all analytes, and there is no single fixed position of the burner which would be suitable for all analytical situations. It is therefore necessary to adjust the burner position for

74

Analysis by atomic spectroscopic methods

[Ch.4

each separate analysis so that the maximum population zone coincides with the optical path. It is also important to remember that altering the fuel-to-oxidant ratio, or changing the total gas flow without altering the stoichiometry can modify the location of the maximum population zone within the flame. Consequently, it is often necessary to readjust the observation height in the flame until maximum absorbance is again obtained. 4.3,23 Flame microsampling In this technique, a very small volume of the sample (typically 50-100 /xl) is injected into a polypropylene or teflon funnel connected to the aspiration capillary of the nebulizerburner system. The small volume of the sample reaches the flame and is atomized. The resulting analytical signal is a fast, transient absorbance peak, which is measured by the peak reading system (peak-height or peak-area mode). Direct injection of samples and calibrants is accomplished using a micropipette of the required volume. With recent instruments, flame microsampling can be performed more easily using the automatic sample dispenser. The major application for microsampling is the determination of relatively high concentrations, or in trace analysis where preconcentration techniques are required. In this case more elements can be determined from a given small volume than by conventional aspiration. Typically, up to ten elements can be determined from 1 ml of sample. Flame microsampling can also be used in many organic solvent applications and high-dissolved solid analyses where conventional aspiration may be troublesome. 4.4

Vapour generation techniques

Analytical requirements for some important elements are often such that the significant analytical level is below the detection limit of conventional flame methods. For some of these elements, special vapour generation techniques can be used to provide the required improvement in the measurable concentration. 4.4.1 Hydride forming elements Antimony, arsenic, bismuth, lead, selenium, tellurium and tin can be determined by chemically reducing the element to gaseous hydride form and then dissociating the hydride in a heated quartz tube placed in the optical path of the spectrometer. Chemically, the procedure is simple. The reduction is performed by adding sodium borohydride pellets or solution to the appropriately acidified sample. The reagent concentrations are chosen to give quantitative evolution of elemental hydride vapour in a few seconds at room temperature. The hydride vapour is then swept by a continuous stream of nitrogen into a quartz tube heated in air-acetylene flame or electrically. The hydride compounds are atomized and the transient absorption signal is measured as a sell-defined peak while the vapour passes through the quartz tube. This derivatization procedure is described in detail for speciation analysis elsewhere in this volume. 4.4.2 Mercury 4,4,2,1 Determination Mercury determinations can be carried out using the described hydride generation system, but the formation of the vapour and the heating of the quartz tube are not required because of the volatility of mercury in the elemental state; hence, it can be directly measured with the cold vapour technique. In this case the atomic absorption

[Ch.4

Analysis by atomic spectroscopic methods

75

Spectrometer is reduced to its absolute minimum: a mercury lamp, a quartz-windowed absorption cell and a specific photodetector at 253.7 nm. The mercury ions present in the acid solution after the mineralization step are therefore reduced to zero valent mercury in a reduction-aeration cell. The use of stannous chloride as reducing agent is preferred to sodium borohydride because a better precision and lower detection level can be achieved. The mercury vapour is then swept out through a drying tube into the quartz cell and the absorption signal is measured. This method was firstly described by Hatch and Ott [24]. For environmental studies, the determination of very small concentration of mercury (below the ng.g'^ level) due to the high toxicity of this metal and its compounds is required. Several preconcentration techniques have been proposed, based on absorption of elemental mercury in liquids or on solid phases. The preconcentration of mercury by amalgamation of a metal surface allows a reversible absorption/desorption; quartz sand coated with gold is a very efficient absorber due to its large active absorbing surface [25], the mercury is liberated by thermal desorption at approximately 800 °Cand introduced into the atomic absorption spectrometer. Quartz tubes filled with gold coated sand can be used for the collection and determination of gaseous mercury in air while mercury compounds associated with particulate matter are retained on a quartz prefilter. 4.4,2.2 Sources of errors Spectral interferences associated with the determination of mercury by cold vapour atomic spectroscopic methods, mostly due to the presence of sulphur dioxide, ammonia or aromatic compounds in the gas stream, can be eliminated by using a gold film mercury detector, based on a linear change in resistivity of a gold film as a function of the degree of amalgamation [26]. Due to the high volatility of mercury and some of its compounds, and its property to adsorb on most surfaces, many systematic errors can occur at the ng.g'Mevel; they are mostly associated with the treatment, storage and mineralization of the samples and are well described by Kaiser et al [27]. Contamination by containers and glassware is often a cause of high background values. 4.5

Hectrothennal atomization

The first electrothermal atomizers for AAS were developed in the early seventies and allowed a considerable improvement in sensitivity. However, this technique suffered from numerous spectral and non spectral interferences which rendered it difficult to handle. A major improvement was made with the introduction by L'vov [28] of a platform into the atomizer to reach near isothermal atomization and the use of efficient chemical modifiers to increase the pyrolysis temperature and thus attain a more specific volatilization of the analyte during the atomization step. These two points made it possible to reduce or totally overcome the influence of most analyzed matrices. 4.5.1 General considerations Electrothermal atomizers have provided a considerable improvement in sensitivity for the majority of the elements determined by atomic absorption. This is due essentially to the introduction of the entire sample into the atomizer, thereby overcoming the inefficiency of nebulization systems and avoiding the large dilution of the atomic population in the flame gases.

76

Ana^is by atomic spectroscopic methods

[Ch.4

Flame and furnace atomizers provide the same end-product: a supply of free analyte atoms for exposure to light at a characteristic wavelength. However, there are major differences in the mechanisms of molecular dissociation and in the overall efficiency of the atom production process in both systems. In flame technology, the chemical composition of the flame, rather than pure temperature, is generally more important in maximizing the free atom fraction. The conversion efficiency is low: typically only about 10 % of the aspirated sample is converted into the free atom population for measurement in the optical path. With electrothermal atomizers, molecular dissociation is governed by the final temperature used, by the rate at which this temperature is attained, and by the reducing environment of the hot graphite. Conversion efficiency is typically high since all of the available sample is used to produce the atom population within the optical path. Thus for a given analyte concentration, the atom population in the furnace atomizer will be considerably more dense than in the flame. Consequently the measured absorbance will be considerably higher. In practical analytical terms, measurements within the useful absorbance range can be obtained at concentrations considerably lower than with flame methods. 4.5.2 Atomizer The graphite tube is mounted in the atomizer head between two graphite electrodes located in the water-cooled metal block. A toggle mechanism locks the graphite tube in place and allows the furnace assembly to be opened for replacement of the tube. The contact area between the electrodes and the tube is relatively small. The tube is thus resistively heated by the passage of a high current at low voltage. The heating (achieved temperatures up to 3000 °Cand ramp rates up to 2000 °C.s"^)of the tube is monitored by an adequate temperature control device. Inert gas (usually argon) flows through the electrodes to protect the inner and outer surfaces of the graphite tube from rapid oxidation by atmospheric oxygen. The graphite tubes are coated with pyrolytic graphite which is, compared to normal graphite, impermeable to solutions and gases. Pyrolytic graphite is also more resistant to oxidation than normal graphite, and is inert. 4.5.3 Electrothermal program For routine analyses, a small volume of sample (typically between 3 and 50 /xl) is introduced into the atomizer through the port at the top of the tube, generally by means of an automatic sample dispenser. The graphite tube is then heated in accordance with a pre-determined electrothermal program. Although there are different approaches in the detailed design and construction of furnace atomizers, they all perform according to the same fundamental process: generating a population of free analyte atoms so that atomic absorption can be measured. In its simplest form, the electrothermal program is achieved in three stages: 1.

The drying step: the solvent is evaporated from the sample at a temperature near the boiling point (± 100 °C and about 2 s per /xl for aqueous solutions). The evaporation must be gentle in order to avoid losses by sputtering. When the drying stage is complete, the residue in the atomizer tube will consist of a thin crust of material, containing the analyte element together with all the solid components of the sample matrix.

[Ch.4

2.

3.

Analysis by atomic spectroscopic methods

77

The ashing (charring, pyrolysis) step which removes the organic or inorganic matrix. This step represent the most important stage of the electrothermal program. The matrix is thermally decomposed at an intermediate temperature (300-1500 °C as a function of the analyte). The objective is to remove the maximum of the matrix while keeping the analyte entirely within the atomizer in a stable form so that atomization can proceed with a minimum interference from the matrix components. During the ashing step, an alternative gas (generally air or oxygen) can be used to achieve a better matrix decomposition. When the ashing step is complete, the residue should consist of the analyte element in its appropriate molecular form, an a minimum of the matrix inorganic salts which are thermally stable at the ashing temperatures used. The atomization step in which the dissociation of the analyte molecular species at a high temperature (up to 3000 °C) occurs, and free analyte atoms are generated within a confined zone coincident with the spectrometer's optical path. The flow of the sheat gas is usually interrupted during the atomization in order to extend the residence time of atoms in the optical path and consequently to enhance the sensitivity of the atomic absorption measurement, performed during this step.

While it is convenient to consider the electrothermal program as a three-stage programming, it is actually divided into a number of small steps. Each step can be separately programmed for duration, rate of temperature rise (ramp parameters) and temperature at the end of the step ( hold parameters). Before performing a real analysis, the ashing and atomization temperatures for an element in a particular matrix must be defined using preliminary programming. Double curves are established, in which the absorbance signal is plotted versus the applied temperature. In the first curve the height of the signal at the fixed atomization temperature (found in the operating conditions supplied by the manufacturer) is plotted versus the ashing temperature as the variable. This curve shows the temperature to which a sample can be thermally pretreated without loss of analyte element, the "optimum ashing temperature". In the second curve, this optimum ashing temperature is fixed and the absorbance signal is plotted versus the now variable temperature of atomization. The temperature at which the maximum atom cloud is attained (maximum absorbance signal) is called the "optimum atomization temperature". In electrothermal A AS, the absorption signal produced during the atomization stage is a transient, well-defined peak. The height and area of the transient absorbance peak from the ETAAS are related to the amount of analyte present in the sample. By analogy to flame AAS, the sensitivity of furnace analysis is called "characteristic mass", mo, and is defined as the amount of the analyte in mass units (pg) that produces 0.0044 absorbance (for peak-height measurements) or 0.0044 absorbance.second (A.s), (for peak-area measurements). When the slope of the working curve is known (in A/pg), the characteristic mass can be calculated as m^ = 0.0044/slope. 4.5,4 Interferences The presence of the sample matrix can lead to interferences. The matrix effect, for example, is a composite interference due to all the concomitants in the sample. The other phenomena (influence of the flame, graphite, sheat gas etc) are not considered as interferences since they are not due to the sample properties.

78

Analysis by atomic spectroscopic methods

[Ch.4

4,5.4,1 Spectral inteiferences Spectral interferences mainly arise from a scattering of the source radiation on nonvolatilized particles from concomitants or from absorption of the source emission line by overlapping atomic lines (in AAS this probability is very low and the problem is well documented) or from the molecular bands of the concomitants. Problems arising from background absorption (light scattering and molecular absorption) are more frequent with electrothermal atomizers than with flames. In chemical flames, background absorption is rarely significant at wavelengths higher than about 240 nm, and the background signal seldom exceeds about 0.05 absorbance. In furnace analysis, background absorbance can be significant at wavelengths up to about 500 nm, and the background signals may exceed 2 absorbance units. Light scattering, which produces a broad band absorption, is caused by the condensation of the sample matrix, after vaporization, by forming a smoke or mist; this event occurs when a huge amount of volatilized matrix reaches the cooler regions at the open ends of the graphite tube. Molecular absorption is caused by the broad absorption bands of molecular species present in the atomizer, particularly alkali-metal and alkalineearth halides. In practice, all these phenomena are observed when analyzing samples with high dissolved salts, mainly halides in sea water, urine etc. Sharp absorption lines, called "structured background" can be superimposed on the broad absorption bands, due to electronic spectra of molecules. To counter the background phenomena, there are three complementary counter measures available to the analyst: simultaneous background correction, appropriate temperature programming, and the use of chemical modifiers. An essential part of an atomic absorption system for many applications, particularly for electrothermal atomization methods, is a device for correcting the atomic absorption signal for non-specific molecular absorption or light scattering effects. Both of these attenuate the light beam, giving an apparently increased absorption signal. A simultaneous background correction device is very convenient in flame work, but absolutely essential in analysis using a furnace. A well-established technique for correction is to measure, in rapid sequence, the radiation attenuation of a primary source which measures the total absorption (the sum of analyte atomic absorption and background absorption), and of a continuum source (background absorption only). A further technique for exact background correction introduced in recent years makes use of the Zeeman effect. Background correction with a continuum source. Systems for background correction using a continuum source are based on the work of Koirtyohann and Pickett [29]. In these devices the radiation from the hollow cathode lamp is passed alternatively (mechanical or electronic chopping, pulsing, modulating) with the radiation of a continuum source through the graphite furnace. The optical configuration is such that radiation from both the hollow cathode lamp and the continuum lamp coincide precisely along the optical path through the observation zone. After passing the monochromator, both radiation beams fall on the same detector and an electronic measuring system forms the ratio from both radiant intensities. The background signal is subtracted electronically from the total absorbance signal and the analytical result is thus corrected for background interference. Most often, the background corrector consists of a deuterium arc emitting a continuum signal (190-360 nm). The deuterium and cathodic radiations are focused

[Ch.4

Analysis by atomic spectroscopic methods

79

in the centre of the furnace with equivalent energies. During the reading, the nonspecific absorptions attenuate the two beams in the same way, while the analyte element absorbs only the cathodic beam because of its narrow line width. It is worthwhile considering at this point how to ensure that the background correction operates with the best possible efficiency. This requires attention to the optical alignment and the chopping frequency. In order to ensure that correction is applied only to the volume within the tube where the atomic absorption is measured, it is essential that the continuum beam be accurately aligned with the atomic beam within the atomizer. The rate at which a cycle of pulses occurs (chopping frequency) is of great importance when the corrector is working with an electrothermal atomizer. As already noted, during the atomization step the atomic population increases quickly to a maximum, then falls more slowly. The background absorption behaves in a similar way, though not identically. The accuracy of correction therefore depends on the time interval (spectrometer time constant) between successive pulses relative to the rate of change of both background and total absorption. The background correction with a continuum source has disadvantages and limitations. In general, it should be attempted to keep the background attenuation below 0.5 absorbance; frequently, higher background values may be incompletely corrected, especially if fast, dynamic signals are generated by the graphite furnace. For samples of unknown composition, it is important to have information of the nature, magnitude and rate of apparition of background signals. For this purpose, most modern atomic absorption spectrometers permit the detailed observation, on a screen display, of both specific and background absorbance signals generated during the atomization step. The spectral continuity of the background signal with the bandpass of the monochromator is an essential condition for a good correction. This is assured for light scattering and dissociation continua. Background correctors using a continuum source are incapable of dealing with background attenuation of electronic excitation spectra, composed of many narrow lines. For this type of spectrum, the actual background correction depends on the degree of overlap between the analyte spectral line and the individual molecular rotational or vibrational lines [30,31]. If a continuum background corrector is used in these conditions, it determines the mean absorbance over the observed spectral bandpass, and this mean absorbance may be higher or lower than the actual attenuation at the analyte wavelength. This results in an over- or an undercompensation when the background readings are subtracted, and thus an invalid analytical measure. From the practical view-point, the problems due to the structured background concern fortunately only the determinations of selenium and arsenic in magnesium and calcium phosphate matrices (plant and animal tissues, soils, blood, urine...). For this applications, the Zeeman background correction is thus strongly recommended. Zeeman effect background correction, A fundamentally different approach to background correction involves the Zeeman effect which occurs when an atomic vapour which is absorbing or emitting resonance radiation is subjected to a magnetic field of several kilogauss. This procedure splits the spectrum lines into a number of components. In the simplest cases, the main resonance line is replaced by a TT component situated at the original wavelength together with two a lines, displaced by equal wavelength intervals at both sides of the original line. Splitting of the spectral lines into three components is designated the normal Zeeman effect and concerns the main resonance lines of Group IIA an IIB elements. All other lines exhibit an anomalous Zeeman effect and split into more than three components.

80

Analysis by atomic spectroscopic methods

[Ch.4

These effects can be used in background correction by applying the magnetic field (permanent or alternating) either to the source of the resonance radiation (direct Zeeman effect, compatible with conventional hollow cathode lamps) or to the atomized sample (inverse Zeeman effect, tht most popular). Besides the Zeeman splitting of the resonance lines, the radiation is also polarized. The polarization varies according the direction of observation and the resulting configurations are called the transverse Zeeman effect (magnetic field perpendicular to the radiation beam) or the longitudinal Zeeman effect (magnetic field parallel to the radiation beam). In most commercial systems, the light from the source is polarized alternatively, by means of a rotating polarizer, in planes perpendicular and parallel to the magnetic field applied to the atomizer, before it actually passes through the atomizer. With the Zeeman effect, only light polarized in a parallel plane is absorbed by the n components (atomic absorption and background attenuation). Light polarized in a perpendicular plane does not undergo any atomic absorption and only submits an attenuation by the background. These components are detected in the usual way. For practical applications it was shown that the optimum Zeeman system has a magnet located at the atomizer (inverse Zeeman effect), operating with an alternating field at a flux density of about 1 Tesla. For practical reasons regarding the instrumentation, the transverse Zeeman effect was first selected. This configuration offers an optimum double-beam system, high accuracy background correction, linearity and sensitivity similar to conventional AAS and detection limits better than can be achieved with a deuterium arc background correction. Such a system permits a correction of background signals up to 2 units of absorbance. The background measurement is performed exactiy at the analyte wavelength: the Zeeman background correction systems can thus deal with structured background, uncorrectable with conventional deuterium devices. Recently, a longitudinal Zeeman system was also introduced. This system theoretically provides a better performance than the transverse configuration, due to a higher energy throughput by omitting the polarizer. 4,5.43 Non-spectral interferences Non-spectral interferences are the major problem when using electrothermal atomizers. They are usually classified according to the place, stage or process responsible for their occurrence, e.g. condensed-phase interferences concern all processes occurring from the formation of compounds and reactions during drying and ashing steps to the complete volatilization of the analyte from the graphite tube. Vapour-phase interferences occur when the analyte is not completely dissociated into atoms in the ground state. Interferences in the condensed phase include all influences on the analyte up to the moment when it volatilizes and leaves the hot graphite surface. They are due to e.g. losses of the analyte during the ashing step, to the formation of carbides or intercalation compounds, to the occlusion of the analyte by the matrix, and similar processes that may result in changes of the analyte volatilization rate or its incomplete atomization. Furthermore, one of the main causes of vapour phase interferences in electrothermal atomizers is the formation of monohalides, principally in the presence of chlorides in the sample (sea water, urine) or in the acid mixture used for the mineralization of the sample (HCl). Most analytes form stable molecules with the halides at the decomposition stage. These molecules are volatilized during the atomization step but they are not dissociated into free atoms. The molecular absorption produced by the undissociated molecules is, evidentiy, corrected as a non-specific absorption, and this part of the analyte is lost for atomic absorption.

[Ch.4

Analysis by atomic spectroscopic methods

81

In comparison with an absorbance signal measured in a simple aqueous medium, a nonspectral interference usually provides a change in the absorbance-time profile obtained in the sample (lowering, enhancement, enlargement etc.). Consequently, the slopes of the working curves established in both simple and complex media differ, and the direct calibration method against aqueous calibrants is not valid. In these cases, the following countermeasures can be of use. Standard addition method. In general, non-spectral interferences can be eliminated by making calibrants as similar as possible to the sample composition. In the ideal case, a calibrant would contain not only the same solvent, but also the same concomitants. Fortunately, interferences are rarely so pronounced that calibrants should match sample solutions exactly. When the sample composition changes markedly or more complex matrix effects are encountered, then the standard additions calibration method is recommended. As a general rule, this procedure should provide a substantially linear calibration since accurate regression cannot be obtained from non-linear calibration points. It is also essential to establish an accurate baseline from the appropriate reagent blank. In several applications of the standard additions method, each sample has to be analyzed individually, e.g. in the case of samples of similar origin but with very different matrices. The most variable matrix from one sample to another can be found with urine samples, for which trace element determination thus becomes particularly difficult. There are some situations in which a batch of samples can be analyzed against the one set of standard additions, but this is only valid when all samples in the batch are chemically and physically similar (e.g. similar matrix from one sample to another: seawater, blood). Using the standard additions method, it should be possible to compensate all non-spectral interferences, provided that they are independent of the analyte concentration. With modern instruments the standard additions method can be performed automatically but it is time-consuming and, hence, less attractive for routine purposes. Atomization off the platform. In the original furnace concept, the graphite tube is heated to a high temperature and the sample is vaporized from the tube wall. This quickly converts the sample into an atomic vapour that absorbs light from the primary source. With the Massman-type furnaces that are currently used, it is not so simple. If a sample is placed on the wall of the graphite tube, the various sample concomitants will vaporize as the wall of the tube heats up, and into an environment which is much cooler than the wall of the tube directly heated. In these conditions, recondensation of the analyte (and sample concomitants) may occur, resulting in a depression of the absorbance signal. However, the temperature and rate at which the volatilization occurs depends on the compound in which the analyte is present. With such differences in volatility, the residence time of analyte atoms in the atomizer will be different as volatilization occurs into an atmosphere that is rapidly increasing in temperature. This is in contradiction with the assumption of the initial L'vov theory that all atoms, regardless of their form, will be volatilized into an atmosphere that is at the same constant temperature. As a solution to this problem, L'vov suggested that the heating of the sample be decoupled as much as possible from the heating of the graphite tube. This was achieved by placing a small, solid pyrolytic graphite platform on the bottom of the graphite tube [28], the sample aliquot being then deposited on this platform. The temperature of the platform, heated primarily by radiation, lags behind that of the tube wall. Hence, vaporization is delayed until the atmosphere reaches a high and nearly constant temperature. The analyte is volatilized from the platform into a gaseous environment which is at higher temperature than the atomization support. This leads to an efficient

82

Ana^is by atomic spectroscopic methods

[Ch.4

decomposition of molecules, minimizes recondensation phenomena and, consequently, drastically reduces vapour-phase interferences. With respect to volatilization from the tube wall, the volatilization from the platform is delayed, and each atom, regardless of the temperature at which it is volatilized, will have the same residence time and contribute equally to the absorption process. From a practic^ point of view, the use of the platforms also extends the useful lifetime of the atomization support. While graphite tubes are coated with only a thin layer of pyrolytic graphite (typically about 50/xm thickness), platforms are made of solid pyrolytic graphite, more resistant to degradation by corrosive agents (acids, oxygen ere.) or samples (seawater, urine etc). Chemical modifiers. Chemical modification techniques, first used by Ediger et al [32], are now widely used in electrothermal atomic absorption spectrometry. They are generally employed in the determination of various volatile and mid-volatile elements, generally atomized off the platform. For the determination of refractory elements, atomized from the tube wall, chemical modification is used more rarely. A chemical reagent (modifier) is added in large excess directly to the calibrants and samples, or to the atomizer before the addition of the analyzed solutions, to convert the analyte into a phase of higher thermostability {analyte modifier), to increase the volatility of the concomitants (m«/nr moJz^er), or for both purposes (analyte/matrix modifier). In the analysis of complex and difficult samples the use of complex (or composite) rather than single-component (individual) chemical modifiers is often preferred. Their various constituents have different roles during the thermal pretreatment of the sample (for example the analyte thermal stabilization and the real modification of the matrix). The action of the usual modifiers is given in Table 1. Analyte modifiers. In most cases, the modifier retains the analyte up to higher temperatures. This, on the one hand, permits matrix components to be removed more easily and more effectively prior to the atomization of the analyte. On the other hand, thermal stabilization of the analyte to higher temperatures results in its volatilization in a hotter vapour phase which is closer to thermal equilibrium. This fact minimizes substantially the risk of vapour-phase interferences. The best example of this kind of modifier is represented by palladium. Palladium is a very effective chemical modifier and can be used to stabilize many elements to several hundred degrees (300-1000 °C) higher than the temperatures possibly obtained with current methods [33]. The greatest temperature shifts are achieved for the following elements: As, Se, Bi, P, Pb, Sb and Tl. The increase of the appearance temperature is attributed to the formation, with palladium, of more refractory intermetallic species or alloys. Other metals of Group VIII (in particular Ni, Ir and Pt) can be also very efficient modifiers in some cases, but the "universality" of palladium is clearly demonstrated.

[Ch.4

Analysis by atomic spectroscopic methods

83

Table 1 - Action of some chemical modifiers

Type of matrix or analyte Matrix with alkali-metal and alkaline-earth halides Ba, Mo, Ti, V, U, Zr

Ag, Cd, Pb As, Au, Bi, Sb, Se, Te Tl P Hg Mn

Al, Be, Cd, Co, Cr Fe, Mn,, Ni,,Pb,, Sn

Modifiers, action Formation of more volatile compounds NH4 nitrate, phosphates ascorbic or oxalic acids Formation of more volatile compounds hydrofluoric acid Formation of more refectory compounds NH4 phosphates, Pd nitrate, Ni, Cu, Pd, Ag, Mo, Fe nitrates, Ir (hexachloroiridate) H2SO4, Pd nitrate Pd, La nitrates Pd nitrate Pt (hexachloroplatinic acid), Pd nitrate Incorporation into a salt or oxide alkali-metal or alkaline-earth nitrates, usually Mg(N03)2

An increase in the volatility of the analyte by means of an analyte modifier can also be used. In this case the modifier acts as a '\olatilizef\ facilitating an atomization at low temperature. These applications, generally using carboxylic acids as modifiers, facilitate the temporal separation of the analyte and background signals. They are employed for the determination of the most volatile elements {e.g. Cd, Zn) in the matrices which generate huge background signals (seawater, urine). Similarly, by using adequate modifiers (generally fluoride-based), the atomization temperature of several refractory elements can be lowered {e.g. Mo, V). Decreasing the atomization temperature has a beneficial effect on the graphite tube lifetime and long-term reproducibility of the determinations of the refractory elements. Matrix modifiers. The role of a matrix modifier is to increase the volatility of interfering concomitants. The matrix can be extracted during the pretreatment stage with a better efficiency. In this case the atomization of the analyte is less disturbed by the possible nonspectral interferences and high level background signals. Their use is not frequent; in practice only ammonium nitrate or nitric acid are largely employed to volatilize chlorides from seawater or urine samples prior to atomization. For a more efficient removal of organic matter from the sample during ashing {e.g. for blood trace-metal analysis), the use of oxygen as an alternative gas can be considered as a matrix modifier. In addition, "ashing aids" have beneficial effects on the organic matrix modifications and simplifications during the thermal decomposition. Magnesium nitrate or nitric acid (generally in combination with another modifier) are mostly used to favour the oxidation processes and to facilitate a more efficient removal of the matrix prior to atomization.

84

Analysis by atomic spectroscopic methods

[Ch.4

Analyte/matrix modifiers. The same chemical modifier acts also on the matrix components during the ashing stage, and modifies their influence during the atomization process of the analyte. For example, in the determination of selenium in a phosphate matrix using a deuterium background correction device, the uncorrected absorbance signal, due to structured background from generated phosphorus species, is shifted to higher temperature values in the presence of palladium (or nickel, iridium etc.) added as a modifier. The analyte absorbance signal is, consequently, not longer superimposed on the background signal and hence, uncorrectable with a conventional deuterium arc device. In these conditions, the selenium absorbance signal can be measured away from the influence of the structured background [34]. In this example, palladium (or nickel) first acts as a thermal stabilizer of selenium, but also modifies the volatilization behaviour of the phosphate matrix. Complex or composite modifiers. The beneficial properties of palladium as a matrix modifier have already been widely demonstrated. However, its use as a single modifier cannot solve all interference problems encountered in furnace analysis of difficult matrices. The efficiency of palladium may still be reinforced in some instances by its combination with one or more other agents. For example, the mixed modifiers composed of palladium and of ammonium or magnesium nitrates were suggested by Welz and coworkers [35,36] for the efficient determination of a number of trace elements in several environmental matrices. Similarly, Beach and Shrader [37] employed a reducedpalladium modifier. The addition of a reducing agent (ascorbic acid, hydroxylamine, hydrochloride or hydrogen added to the sheat gas) ensured a more consistent performance. The way is now open to elaborate complex modifiers. The easiest way, in our opinion, would be to use matrix interferents themselves to complete the action of an analyte modifier in order to simulate, as far as possible, the same analytical conditions encountered in both simple and complex media (calibrants and samples). Some of the main sample matrix elements act as interfering agents. In the presence of a basic analyte modifier (e.g. palladium), it is assumed that by adding these interfering elements in large excess to the samples and calibrants, the interferences produced by the samples themselves will be negligible. This assumption appeared to be valid for numerous types of furnace analyses of various environmental samples [38]. The following example describes the step-by-step procedure carried out to ensure an interference-free determination of lead in various environmental samples. In a simple aqueous medium, the ashing step for lead should not exceed 500 °Cas, above this temperature, losses of the analyte will occur. When analyzing a complex sample under these conditions, the matrix is generally not sufficiently removed during the ashing stage, which results in a disturbed atomization of lead. With the addition of palladium, the charring temperature may be elevated to up to about 1200 °C,a temperature sufficiently high to remove an important part of the matrix. However, in the case of the analysis of simple aqueous calibrant, the lead atomization rate (peak-height) decreases (the palladium only acts as an interfering element in this case), whereas, in the analysis of environmental matrices (plant and animal tissues) the lead atomization process is largely accelerated (enhanced peak-height signal). This clearly shows that, in the presence of palladium, one or more matrix components contribute to this enhancement which is not observed in the simple aqueous medium. Additional experiments have shown that the different behaviour of lead during the atomization is due to the presence of phosphorus and magnesium, two main matrix elements of most environmental matrices; a mixed modifier composed of

[Ch.4

Analysis by atomic spectroscopic methods

85

palladium and of excess ammonium phosphate and magnesium nitrate may therefore be succesfuUy used: palladium ensures the thermal stabilization of lead, phosphates accelerate its atomization process (by enhancing its peak-height signal), and magnesium nitrate acts as an ashing aid and allows normalization of the differences in the absorbance peak-shapes observed for the various matrices. Consequently, with such a mixed modifier, charring may be performed at 1100 ''C,and the lead absorbance-time profiles for all the matrices studied, including simple aqueous medium, have similar shapes. This is verified by a similarity in the slopes of standard addition curves established in the different matrices, indicating the absence of non-spectral interferences. In this case, a direct calibration against aqueous calibrants is valid for matrices as different as soils, sediments, suspended matter samples, animal and plant tissues or sea water [39,40]. This example permits the evaluatation of the great potential of chemical modifiers, and shows the probable fields of future research into the analytical applications of electrothermal atomic absorption spectrometry. Another example is given by the determination of cadmium in seawater. Cadmium is one of the elements with a very high sensitivity (« 0.4 pg for 0.044 absorbance). In a simple aqueous medium, the pyrolysis temperature should not exceed 380 °C. This temperature can be increased in the presence of nitric acid [38] or other compounds as phosphates [41] but losses of cadmium can be expected at 600-700 °C. In saline samples, the cadmium and salt matrix are volatilized at the same temperature: both the analyte and background absorbance signals hence appear simultaneously which leads to difficulties in the cadmium determination. In seawater samples these background signals are too high to be satisfactorily controlled by deuterium or Zeeman correction devices. The use of platinum metals-based modifiers offers a possibility of substantial increase of pyrolysis temperature for several elements, e,g. Pb, Mn, Tl, As, Se, Cu, Te... [38] Unfortunately, the determination of cadmium cannot benefit from this advantage. With palladium, a pyrolysis temperature higher than 700-800 °C cannot be used without provocating losses of cadmium. Such a temperature is generally sufficient for the analysis of matrices such as water, plant and animal tissues, soils and sediments, but for seawater analysis it is not enough to remove a significant part of the matrix during the pyrolysis step as uncorrectable background levels will be generated during the atomization step. Cadmium atomization at relatively low temperature, before the appearance of high background levels, requires a suitable temperature program: firstly the unnecessary charring step, which is in any case conducted at a temperature too low to remove the seawater matrix, is skipped. The drying of the sample is then directly followed by the atomization step. Possible losses of cadmium due to inadequate charring temperatures are thus totally avoided; secondly, cadmium is atomized at a temperature as low as possible to avoid the huge volatilization of the matrix and hence high background levels. The best results are obtained at 1400 °C,a temperature at which the atomization rate of cadmium is sufficient and the background signal is satisfactorily controlled by the correction device. The analyte signal is not directly subject to interference by the increasing background signal (as observed at a temperature up to 1500 °C) and the integration time is limited. The salts remaining in the atomizer after this low temperature atomization step are removed during an additional cleaning step conducted at 2400 ""C. The situation is still improved by using an argon-hydrogen mixture as the sheat gas. In this case, the cadmium absorbance signal is entirely out of the background signal influence, and the limitation of the integration time is not necessary anymore [42].

86

Analysis by atomic spectroscopic methods

[Ch.4

4.5.5 Role of the graphite Since the commercialization, in the early seventies, of graphite furnace atomizers, concern has arisen about the variability of the surface properties of graphite tubes and their progressive degradation with increasing numbers of firing. Although the use of spectrographic graphite has specific advantages, such as a mechanical resistance increasing with temperature up to 2500 °C and good reduction properties, several drawbacks exist. Firstly, graphite can interact with some elements to produce stable carbides and/or interlamellar compounds. Both of these phenomena may hamper the volatilization of analyte and change its original atomization process. Secondly, graphite is porous and the atomizer wall is therefore subject to infiltration by solutions. In addition, vapour can diffuse at high temperature through the atomizer wall. Coating of the tube wall with pyrolytic graphite, which is non-porous and highly impermeable to gases at the temperatures generally applied, has greatiy reduced these problems [43,44]. However, the thin pyrolytic graphite layer itself becomes increasingly porous with tube aging as a result of the influence of strongly acidic solutions, corrosive samples and high atomization temperatures. Other coatings and impregnation materials have been studied in order to improve the atomizer surface properties. Metals with an elevated melting point and forming stable interstitial carbides with graphite (Ta, Mo, Nb and W) have often been used for this purpose [45,51]. However, improvements obtained by these treatments are too case-specific to be universally applicable. Thirdly, the quality of the pyrolytic coating may be highly variable and is a function of the number of active sites per unit area. This number depends on the initial number of nucleation centres available on the graphite surface during the coating process [52]. Oxygen provided by the injected sample solution will bind to carbon preferentially on these active sites. During atomization, this carbon is released in the vapour phase as CO gas, increasing the active site oxygen-binding capacity. Also, analyte and matrix elements will compete for these active sites. Such processes are variable during the tube lifetime and can favour or inhibit certain reactions [53]. They can also modify the initial process of analyte atomization. In practice, it is extremely difficult to distinguish between the different processes involved and to identify the ones which are controlled mainly by active site availability [54]. The situation described concerns mainly the determination of refractory elements. The determination of more volatile elements is less affected by these problems. 4.6

Atomic absorption spectrometry /conclusions

Flame atomic absorption spectrometry is now very well established, relatively inexpensive and easy to use: it requires little operator experience. The interferences are rare, well known and easily controllable. Only the refractory elements (V, Mo, Ta, W etc) are not completely dissociated in the flame and are therefore not easily determined. Also elements that have their analytical lines at low wavelengths (As, Se, Te etc) exhibit a sensitivity that is insufficient for the usual analytical problems. If the lowest detection limits that can be obtained with current spectroscopic methods are required, electrothermal atomization is the technique of choice. On a relative basis, the detection limits obtained with a graphite furnace are 10-100 times better than with a flame. In the recent past, however, numerous interferences were reported and the use of this technique in routine work was practically impossible in many cases. These problems are now largely controlled using a combination of platforms, chemical modifiers, efficient background correction devices and modern spectrometric instrumentation. However, furnace

[Ch.4

Analysis by atomic spectroscopic methods

87

determinations remain slow - typically several minutes per element and sample. Also the analytical range is not very large, about 1-3 orders of magnitude. Thus, the electrothermal atomization must be employed only when the flame provides insufficient detection limits.

REFERENCES [I] [2] [3] [4] [5] [6] [7] [8] [9] [10] [II] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

T.T. Gorsuch, ''Destruction of Organic MaUef\ Pergamon Press, Oxford, (1970) Ph. Quevauviller, J.L. Imbert and M. Olle, Microchim, Acta, in press F.J. Langmyhr and G. Wibetoe, Prog, Anal At Spectrose, 8, 1 (1985) U. Volkopf, Z. Grobensky, K. Tamm and B. Welz, Analyst, 110, 573, (1985) Z.A. Khammas, J. Marshall, D. Littlejohn, J.M. Ottaway and S.C. Stephen, Michrochim, Acta, 1, 333 (1985) A. Janssen, B. Bruckner, K.H. Grobecherker and V. Kurfiirst, Fresenius Z Anal. Chem.,322,713 (19S5) A. A. Brown, G. Kiilmer and M. Lee, Philips Analytical Journal, Cambridge (UK) (1987) K.P. Schmidt and H. Falk, Spectrochim, Acta, 42B, 431 (1987) D.V. Brady, J.G. Montalvo, J. Jung and R.A. Curran, Atom, Absorpt Newsl. 13, 5 (1974) D.V. Brady, J.G. Montalvo, G. Joseph, G. Glowacki and A. Pisciotta, Anal Chim, Acta, 70, 448 (1974) R. Karwowska and K.W. Jackson, Spectrochim, Acta, 41B, 947 (1986) L. Ebdon and E.H. Evans, /. Anal Atom, Spectrom, 2, 39 (1987) L. Ebdon and J.R. Wilkinson, /. Anal Atom, Spectrom, 2, 39 (1987) L. Ebdon and H.G.M. Parry, /. Anal Atom, Spectrom, 2, 131 (1987) K.X. Jackson, M.W. Hinds and R. Karwowska, paper presented at CASALS 86, Toronto, Canada (1986) M. Hoenig and P. Van Hoeyweghen, Anal Chem,,5%, 2614 (1986) D. Littlejohn, S.C. Stephen and J.M. Ottaway paper presented at SAC 8613rd BNASS, Bristol, UK (1986) M. Hoenig, P. Regnier and R. Wollast, /. Anal Atom. Spectrom., A, 631 (1989) M. Pinta et le C.1,1. Analysis, 3, 345 (1975) M. Hoenig and R. de Borger, Spectrochim, Acta, 33B, 873 (1983) M. Hoenig and A. Cilissen, Spectrochim. Acta, 48B, 1303 (1993) E. Walsh, Spectrochim, Acta, 7, 108 (1955) J.B. Willis, Nature, 207, 715 (1965) W.R. Hatch and W.L. Ott, Anal Chem,,40, 2085 (1968) R. Dumarey, R. Heindryckx, R. Dams and J. Hoste, Anal Chim, Acta, 107, 159 (1979) P.J. Murphy, Anal Chem,,51, 1599 (1979) G. Kaiser, D. Goetz, G. Toelg, G. Knapp, B. Maichin and H. Spitzy, Fresenius Z. Anal Chem.,291, 278 (1978) B.V. L\ow,Spectochim. Acta, 33B, 153 (1978) S.R. Koirtyohann and E.E. Pickett, Anal Chem,, 37, 60 (1965) C. Hendrikx-Jongerius and L. De Galan, Anal Chim, Acta, 87, 259 (1976)

Ana^is by atomic spectroscopic methods

[31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]

[Ch.4

H. Massmann, Z. El Gobhary and S. Giicer, Spectrochim, Acta, 31B, 399 (1976) R.D. Ediger, G.E. Peterson and J.D. Kerber, Atom, Absopt NewsL, 13, 61 (1974) X-Q. Shan and Z.M. Ni,Acta Chim, Sin.,31, 575 (1981) M. Hoenig and P. Van Hoeyweghen, Int, J. Environ. Anal. Chem.,24, 193 (1986) X.R. Yin, G. Schlemmer and B. Welz, Anal. Chem.,59, 1462 (1987) B. Welz, G. Schlemmer and J.R. Modakavi, paper presented at XXV SCI, Toronto, (1987) L.M. Voth-Beach and D.E. Schrader, /. Anal. Atom. Spectrom.,2, 45 (1987) M. Hoenig and A.M. De Kersabiec, "L'atomisation Electrothermique en Spectrom^trie dAbsorption Atomique'\ Masson Miteurs, Paris, (1990) P. Regnier, R. WoUast and M. Hoenig, ICES Bulletin, in press M. Hoenig, Analusis, 16, LXXVI (1988) M. Hoenig, R. Vanderstappen and P. Van Hoeyweghen, Analusis, 7, 17 (1979) M. Hoenig, E. Puskaric, P. Choisy and M. Wartel, Analusis, 19, 285 (1991) J. Aspilak, C.L. Chakrabarti and A.P. Bratzel, Anal. Chem.,U, 1718 (1972) T. Kantor, S.A. Clybum and C. Veillon, Anal. Chem.,46, 2205 (1974) R. Cioni, A. Mazzucotelli and G. Ottonello, Anal. Chim. Acta, 82, 415 (1976) I.A. Kuzovlev, Y.N. Kuznetsov and O.A. Sverdlina, Zavod. Lab. 39, 428 (1973) H.M. Ortner and E. Kantuscher, Talanta, 22, 581 (1975) J.H. Runnels, R. Merryfield and H.B. Fisher, Anal. Chem.,41, 1258 (1975) T. Stiefel, K. Schultze, G. Tolg and H. Zom, Anal. Chim. Acta, 87, 67 (1976) J.F. Zatka, Anal. Chem.,50, 538 (1978) E. Norval, H.G.C. Human and L.R.P. Buttler, Anal. Chem.,51, 2045 (1979) I. Rubeska, personal communication M. Hoenig, F. Dehairs and A.M. Kersabiec, /. Anal Atom. Spectrom., 1, 449 (1986) J. Koredkova, W. Freeh, E. Lundberg, J.A.Person and A. Cedergren, Anal. Chim. Acta, 130, 267 (1981)